The duocarmycins, first isolated from a culture broth of Streptomyces species, are parent members of a family of antitumor antibiotics that also includes CC-1065. These extremely potent agents allegedly derive their biological activity from an ability to sequence-selectively alkylate DNA at the N3 of adenine in the minor groove, which initiates a cascade of events that terminates in an apoptotic cell death mechanism.1 
Although CC-1065 has shown very potent cytotoxicity, it could not be used in the clinic because of serious delayed hepatotoxicity.2 This observation led to the development of synthetic analogs of CC-1065 (see for CC-1065 derivatives for example Aristoff et al., J. Org. Chem. 1992, 57, 6234; Boger et al., Bioorg. Med. Chem. Lett. 1996, 6, 2207; Boger et al., Chem. Rev. 1997, 97, 787; Milbank et al., J. Med. Chem. 1999, 42, 649; Atwell et al., J. Med. Chem. 1999, 42, 3400; Wang et al., J. Med. Chem. 2000, 43, 1541; Boger et al., Bioorg. Med. Chem. Lett 2001, 11, 2021; Parrish et al., Bioorg. Med. Chem. 2003, 11, 3815; Daniell et al., Bioorg. Med. Chem. Lett. 2005, 15, 177; Tichenor et al., J. Am. Chem. Soc. 2006, 128, 15683; Purnell et al., Bioorg. Med. Chem. 2006, 16, 5677; EP0154445; WO 88/04659; WO 90/02746; WO 97/12862; WO 97/32850; WO 97/45411; WO 98/52925; WO 99/19298; WO 01/83482; WO 02/067937; WO 02/067930; WO 02/068412; WO 03/022806; WO 2004/101767; and WO 2006/043839), which generally showed to have similar cytotoxicity, but reduced hepatotoxicity. Still, however, these derivatives lack sufficient selectivity for tumor cells as the selectivity of these agents—and cytotoxic agents in general—is for a considerable part based on the difference in the rate of proliferation of tumor cells and normal cells, and therefore they also affect healthy cells that show a high proliferation rate. This typically leads to severe side effects. Drug concentrations that would completely eradicate the tumor cannot be reached because of dose-limiting side effects such as gastrointestinal tract and bone marrow toxicity. In addition, tumors can develop resistance against anticancer agents after prolonged treatment. In modern drug development, targeting of cytotoxic drugs to the tumor site can be considered one of the primary goals.
One promising approach to obtain selectivity for tumor cells or tumor tissue is to exploit the existence of tumor-associated antigens, receptors, and other receptive moieties, which can serve as a target. Such a target may be upregulated or to some degree be specifically present in tumor tissue or in closely associated tissue, such as neovascular tissue, with respect to other tissues in order to achieve efficient targeting. Many targets have been identified and validated and several methods to identify and validate targets have been developed.3 
By coupling a ligand, e.g. an antibody or antibody fragment, for such a tumor-associated antigen, receptor, or other receptive moiety to a therapeutic agent, this agent can be selectively targeted to tumor tissue.
Another promising approach to obtain selectivity for tumor cells or tumor tissue is to exploit the existence of tumor-associated enzymes. A relatively high level of tumor-specific enzyme can convert a pharmacologically inactive prodrug, which consists of an enzyme substrate directly or indirectly linked to the toxic drug, to the corresponding drug in the vicinity of or inside the tumor. Via this concept a high concentration of toxic anticancer agent can be selectively generated at the tumor site. All tumor cells may be killed if the dose is sufficiently high, which may decrease development of drug-resistant tumor cells.
Enzymes have also been transported to the vicinity of or inside target cells or target tissue via for example antibody-directed enzyme prodrug therapy (ADEPT)4, polymer-directed enzyme prodrug therapy (PDEPT) or macromolecular-directed enzyme prodrug therapy (MDEPT)5, virus-directed enzyme prodrug therapy (VDEPT)6, or gene-directed enzyme prodrug therapy (GDEPT)7. With ADEPT, for example, a non-toxic prodrug is selectively converted into a cytotoxic compound at the surface of target cells by an antibody-enzyme conjugate that has been pretargeted to the surface of those cells.
Yet another promising approach to obtain selectivity for tumor cells or tumor tissue is to exploit the enhanced permeability and retention (EPR) effect. Through this EPR effect, macromolecules passively accumulate in solid tumors as a consequence of the disorganized pathology of angiogenic tumor vasculature with its discontinuous endothelium, leading to hyperpermeability to large macromolecules, and the lack of effective tumor lymphatic drainage.8 
By coupling a therapeutic agent directly or indirectly to a macromolecule, said agent can be selectively targeted to tumor tissue.
Besides efficient targeting, another criterion for the successful application of targeted conjugates of cytotoxic agents in tumor therapy is that the one or more agents are released efficiently from the conjugate. A further important criterion is that the conjugate is non-cytotoxic or only very weakly cytotoxic, whereas the cytotoxic agent itself exhibits highly potent cytotoxicity. Again another important criterion is that the conjugate must have suitable pharmacological properties, such as sufficient stability hi the circulation, low aggregation tendency, and good water solubility.
Several conjugates of CC-1065 and derivatives have been described (see for conjugates of CC-1065 derivatives for example Suzawa et al., Bioorg. Med. Chem. 2000, 8, 2175; Jeffrey et al., J. Med. Chem. 2005, 48, 1344; Wang et al., Bioorg. Med Chem. 2006, 14, 7854; WO 91/16324; WO 94/04535; WO 95/31971; U.S. Pat. No. 5,475,092; U.S. Pat. No. 5,585,499; U.S. Pat. No. 5,646,298; WO 97/07097; WO 97/44000; U.S. Pat. No. 5,739,350; WO 98/11101; WO 98/25898; U.S. Pat. No. 5,843,937; U.S. Pat. No. 5,846,545; WO 02/059122; WO 02/30894; WO 03/086318; WO 2005/103040; WO 2005/112919; WO 2006/002895; and WO 2006/110476). These conjugates do not possess all of the favorable properties mentioned above. As an illustrative example, glycoside conjugates of seco CC-1065 analogs (analogs in which the cyclopropyl ring as present in CC-1065 is “opened”) have been described that can be activated at the lesion site via an ADEPT approach.9 The difference in cytotoxicity between the conjugates and the corresponding drugs (the cytotoxicity quotient, herein defined as IC50, conjugate/IC50, parent drug) was however relatively low, and the seco CC-1065 analogs themselves did not show extremely potent cytotoxicity. Improvements to the cytotoxicity quotient were made through the development of glycoside conjugates of seco CC-1065 derivatives with a secondary leaving group.10 Although these conjugates demonstrated a high cytotoxicity quotient, their pharmacological properties were non-optimal. For instance, they showed poor water solubility, a consequence of the inherent lipophilic nature of the CC-1065 class of compounds.
Accordingly, there is a clear need in the art for conjugates of CC-1065 derivatives that show high cytotoxicity quotients, contain CC-1065 derivatives that have potent cytotoxicity and favorable pharmacological properties, and release the CC-1065 derivatives effectively.